[mythtv] Ticket #2817: Unknown problem with socket communications causes segfault of backends

Tue Jan 16 23:27:03 UTC 2007

Okay, here's the latest:

Now running revision 12497 / 12490 (master running 12497 due to dts.h bug)

Everything has been fine over the weekend where I didn't have any directories assigned to 'High Quality' on the master... Maybe the entire group vanished... Not sure.

Anyway... Played a few shows today before mounting /vol/TV3. No problems whatsoever.

Mounted /vol/TV3 but didn't assign it to a storage group. Still no problem. Good.

Ran mythtv-setup on my master backend and put just /vol/TV3 as directory for 'High Quality' after creating the 'High Quality' group...

Went to view a show. Got a segfault the first time I tried seeking in the file. mythallways got it back up again in no time. Seeking afterwards was fine. The frontend didn't blackout this time.
Next show - played back just fine, no seeking issues. Exited the show and pressed 'u' to see which host it was recorded on. segfault. Interestingly, it is recorded on the third backend (the one with /vol/TV3 proper). mythallways got the master back up again. the frontend came back to life.
Played third show. Recorded on the backend the frontend is running from (the second). No problems this time (quick seeking and exiting)
Next show - marked start and end, exited, made it transcode, no problems. Recorded on the master backend...
Fith one - nothing to report.
Sixth - also no problems...
Seventh - segfault seconds after seeking began. Actually there was a segfault before that that I didn't even notice...?! Second time - no problem. Recorded on third backend.

I'll now try to set the 'High Quality' Storage Group to /vol/TV on the master in stead..

Got this when I exited mythtv-setup:
2007-01-16 23:36:38.613 Connected to database 'mythconverg' at host: 192.168.33.12
2007-01-16 23:36:59.530 Connecting to backend server: 192.168.33.12:6543 (try 1 of 5)
2007-01-16 23:37:06.541 MythSocket(8230f38:12): readStringList: Error, timeout (quick).
2007-01-16 23:37:06.707 Unexpected response to MYTH_PROTO_VERSION:

A fith segfault occured at some point...

No new segfaults as of yet, but the first show I tried to watch did give me a black screen.. I'll wait a bit for some of the transcode jobs to finish...
Load at around 10 right now.... I think it's the same transcode-job running three times!! Either that or it has three threads... Okay, not three threads. The segfaults must have started new instances... One finished, killed the other two.

One more segfault have occured, maybe when I killed the two mythtranscode processes.

Eighth show - segfault again... It doesn't seem to be NFS-related, then... The frontend when black.

Removed the 'High Quality' Storage Group on the master. Restarted the frontend.

Now the eithth show will start.

Bummer. Just when I though I was in the clear:

/usr/local/bin/mythallways: line 7: 32701 Lagersegmentfejl        /usr/local/bin/mythbackend -l /var/log/mythbackend.log
*** glibc detected *** /usr/local/bin/mythbackend: malloc(): memory corruption (fast): 0x08510da7 ***
======= Backtrace: =========
/lib/libc.so.6[0xb5de5911]
/lib/libc.so.6[0xb5de7e10]
/lib/libc.so.6(malloc+0x86)[0xb5de9206]
/usr/lib/libstdc++.so.6(_Znwj+0x27)[0xb5f86327]
/usr/local/bin/mythbackend(_ZN10QValueListI7QStringE14detachInternalEv+0x3c)[0x806bcac]
[0xbf926e98]
======= Memory map: ========
08048000-08163000 r-xp 00000000 08:09 195556     /home/local/bin/mythbackend
08163000-08164000 rw-p 0011b000 08:09 195556     /home/local/bin/mythbackend
08164000-0859b000 rw-p 08164000 00:00 0          [heap]
ad300000-ad321000 rw-p ad300000 00:00 0
ad321000-ad400000 ---p ad321000 00:00 0
ad4ff000-ad500000 ---p ad4ff000 00:00 0
ad500000-add00000 rwxp ad500000 00:00 0
add00000-aded7000 rw-p add00000 00:00 0
aded7000-adf00000 ---p aded7000 00:00 0
adf00000-ae000000 rw-p adf00000 00:00 0
ae010000-ae03b000 r-xp 00000000 08:08 27072      /usr/lib/liblcms.so.1.0.15
ae03b000-ae03d000 rw-p 0002a000 08:08 27072      /usr/lib/liblcms.so.1.0.15
ae03d000-ae03f000 rw-p ae03d000 00:00 0
ae03f000-ae0ac000 r-xp 00000000 08:08 30191      /usr/lib/libmng.so.1.1.0.9
ae0ac000-ae0af000 rw-p 0006c000 08:08 30191      /usr/lib/libmng.so.1.1.0.9
ae0af000-ae0cd000 r-xp 00000000 08:08 23250      /usr/lib/libjpeg.so.62.0.0
ae0cd000-ae0ce000 rw-p 0001d000 08:08 23250      /usr/lib/libjpeg.so.62.0.0
ae0f1000-ae0f6000 r-xp 00000000 08:08 39786      /usr/lib/qt3/plugins/imageformats/libqjpeg.so
ae0f6000-ae0f7000 rw-p 00005000 08:08 39786      /usr/lib/qt3/plugins/imageformats/libqjpeg.so
ae0f7000-ae0f8000 ---p ae0f7000 00:00 0
ae0f8000-ae8f8000 rwxp ae0f8000 00:00 0
ae8f8000-ae8f9000 ---p ae8f8000 00:00 0
ae8f9000-af0f9000 rwxp ae8f9000 00:00 0
af0f9000-af0fa000 ---p af0f9000 00:00 0
af0fa000-af8fa000 rwxp af0fa000 00:00 0
af8fa000-af8fb000 ---p af8fa000 00:00 0
af8fb000-b00fb000 rwxp af8fb000 00:00 0
b00fb000-b00fc000 ---p b00fb000 00:00 0
b00fc000-b08fc000 rwxp b00fc000 00:00 0
b08fc000-b08fd000 ---p b08fc000 00:00 0
b08fd000-b10fd000 rwxp b08fd000 00:00 0
b10fd000-b10fe000 ---p b10fd000 00:00 0
b10fe000-b18fe000 rwxp b10fe000 00:00 0
b18fe000-b18ff000 ---p b18fe000 00:00 0
b18ff000-b20ff000 rwxp b18ff000 00:00 0
b20ff000-b2100000 ---p b20ff000 00:00 0
b2100000-b2900000 rwxp b2100000 00:00 0
b2900000-b2a00000 rw-p b2900000 00:00 0
b2a01000-b2a05000 r-xp 00000000 08:08 181714     /usr/lib/qt3/plugins/imageformats/libqmng.so
b2a05000-b2a06000 rw-p 00003000 08:08 181714     /usr/lib/qt3/plugins/imageformats/libqmng.so
b2a06000-b2a08000 r-xp 00000000 08:09 195633     /home/local/lib/mythtv/filters/libquickdnr.so
b2a08000-b2a09000 rw-p 00002000 08:09 195633     /home/local/lib/mythtv/filters/libquickdnr.so
b2a09000-b2a0a000 r-xp 00000000 08:09 195634     /home/local/lib/mythtv/filters/libonefield.so
b2a0a000-b2a0b000 rw-p 00000000 08:09 195634     /home/local/lib/mythtv/filters/libonefield.so
b2a0b000-b2a0c000 r-xp 00000000 08:09 195635     /home/local/lib/mythtv/filters/liblinearblend.so
b2a0c000-b2a0d000 rw-p 00001000 08:09 195635     /home/local/lib/mythtv/filters/liblinearblend.so
b2a0d000-b2a0f000 r-xp 00000000 08:09 195643     /home/local/lib/mythtv/filters/libkerneldeint.so
b2a0f000-b2a10000 rw-p 00001000 08:09 195643     /home/local/lib/mythtv/filters/libkerneldeint.so
b2a10000-b2a14000 r-xp 00000000 08:09 195639     /home/local/lib/mythtv/filters/libivtc.so
b2a14000-b2a15000 rw-p 00003000 08:09 195639     /home/local/lib/mythtv/filters/libivtc.so
b2a15000-b2a16000 r-xp 00000000 08:09 195638     /home/local/lib/mythtv/filters/libinvert.so
b2a16000-b2a17000 rw-p 00000000 08:09 195638     /home/local/lib/mythtv/filters/libinvert.so
b2a17000-b2a18000 r-xp 00000000 08:09 195644     /home/local/lib/mythtv/filters/libforce.so
b2a18000-b2a19000 rw-p 00000000 08:09 195644     /home/local/lib/mythtv/filters/libforce.so
b2a19000-b2a1b000 r-xp 00000000 08:09 195641     /home/local/lib/mythtv/filters/libdenoise3d.so
b2a1b000-b2a1c000 rw-p 00001000 08:09 195641     /home/local/lib/mythtv/filters/libdenoise3d.so
b2a1c000-b2a1d000 ---p b2a1c000 00:00 0
b2a1d000-b321d000 rwx/usr/local/bin/mythallways: line 7:   888 Afbrudt (SIGABRT)       /usr/local/bin/mythbackend -l /var/log/mythbackend.log

It does seem that the amount of segfaults lessen significantly if I only have the 'default' Storage Group, though....

/Martin.

----- Martin Møller <martin at martinm-76.dk> wrote:
> I'm not quite sure yet, but I think many of my segfaults was a result
> of Storage Groups.
> 
> I had defined a 'High Quality' storage group that had these
> directories in it on the master:
> /vol/TV3
> /vol/TV
> 
> On my third backend it had this:
> /vol/TV3
> 
> Nothing on the second backend.
> 
> /vol/TV3 was NFS-mapped on the Master.
> 
> I noticed by chance that when /vol/TV3 was *NOT* mounted, I didn't
> seem to get the crashes anymore so now the question is whether the
> issue is because there were two directories or because /vol/TV3 was
> NFS-mapped.
> 
> The directory had read/write rights and root was not squashed.
> 
> I might try to mount it via samba in the coming days and see what
> happens. Hope this can help in finding out what's going on.
> 
> /Martin.